© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 48323
    Prev
    Next
    INDEX
    Explanations

    header guards or 'h' preceding delimiters

    np_acts-logits-general · gemini-2.5-flash-lite

    This neuron fires on short, vowel‐less tokens—typically abbreviations or acronyms (e.g. “hr,” “ssh,” “HF,” single-letter codes) rather than ordinary words.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    .”
    -2.63
     また
    -2.55
    以上です
    -2.48
     was
    -2.44
    它的
    -2.34
     unquestionably
    -2.31
    ſelves
    -2.30
    </h3>
    -2.28
    u
    -2.27
    噥
    -2.27
    POSITIVE LOGITS
    </b>
    2.64
     essentials
    2.55
    {
    2.52
    UAWEI
    2.44
     serían
    2.41
    去
    2.36
     semblent
    2.31
    ated
    2.30
     theses
    2.22
     hig
    2.16
    Activations Density 3.251%

    No Known Activations